A non-invasive blood glucose detection system, method, device and storage medium

Abstract

This application discloses a non-invasive blood glucose detection system, method, device, and storage medium. The system includes: an optical path module, a circuit module, an interaction module, and a control module. One end of the optical path module is connected to the control module via the circuit module, and the other end of the optical path module is connected to the control module. The control module is connected to the interaction module. The optical path module is used to illuminate a preset collection point on the object with light emitted from a light emitter, and then determine photoelectric signal information through photoelectric detection. The interaction module is used to receive basic information about the object. The control module includes a neural network model, which is used to detect the object's blood glucose based on the photoelectric signal information and the basic information. This system can detect the object's blood glucose and improves the accuracy of blood glucose detection. Furthermore, the system is simple to operate, which improves the user experience. This application can be widely applied in the field of artificial intelligence technology.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and in particular to a non-invasive blood glucose detection system, method, device and storage medium. Background Technology

[0002] Patients with related diseases need to frequently test their blood glucose levels. However, among the related technologies, venous blood sampling and electrochemical blood glucose testing are all invasive or minimally invasive methods, which can easily increase the risk of infection for patients. As a result, non-invasive blood glucose testing methods such as blood replacement therapy, biosensor therapy, and energy metabolism conservation therapy have emerged. However, all of these non-invasive blood glucose testing methods have the problems of complex testing devices and low accuracy of test results. Summary of the Invention

[0003] The purpose of this application is to at least partially solve one of the technical problems existing in the prior art.

[0004] Therefore, the purpose of this invention is to provide a high-precision non-invasive blood glucose detection system, method, device, and storage medium.

[0005] To achieve the above-mentioned technical objectives, the technical solutions adopted in the embodiments of this application include:

[0006] On one hand, embodiments of this application provide a non-invasive blood glucose detection system, including:

[0007] The non-invasive blood glucose detection system of this application includes: an optical path module, a circuit module, an interaction module, and a control module; one end of the optical path module is connected to the control module through the circuit module, the other end of the optical path module is connected to the control module, and the control module is connected to the interaction module; wherein, the optical path module includes a light emitter, a photodetector, and a detection device; the light emitter includes a first light source and a second light source, the first light source and the second light source being used to emit two light signals of different wavelengths; the detection device is used to approach a preset collection point on the object to perform blood glucose detection; the photodetector includes a first detector, a second detector, and a third detector, the first detector being used to detect the first light... The system comprises a first light source and a second light source, a second detector for detecting the transmitted or reflected photoelectric signals of the first light source, and a third detector for detecting the transmitted or reflected photoelectric signals of the second light source. The optical path module is used to determine the photoelectric signal information by photoelectric detection after the light emitted by the light emitter illuminates a preset collection area of ​​the object. The circuit module includes a time-division multiplexing unit for implementing time-division multiplexing of the first and second light sources. The interaction module receives basic information about the object. The control module includes a neural network model for detecting blood glucose levels in the object based on the photoelectric signal information and the basic information. This system, by inputting the collected photoelectric signal information from the first and second light sources into the neural network model, can detect blood glucose levels in the object, improving the accuracy of blood glucose detection. Furthermore, the system is simple to operate, enhancing the user experience.

[0008] In addition, the non-invasive blood glucose detection system according to the above embodiments of this application may also have the following additional technical features:

[0009] Furthermore, in the non-invasive blood glucose detection system of this application embodiment, the neural network model includes a data input unit, a preprocessing unit, a first neural network unit, a second neural network unit, and a first output unit;

[0010] The data input unit is used to input the photoelectric signal information and the basic information, and the preprocessing unit is used to filter and normalize the photoelectric signal information to obtain the filtered signal and the difference signal.

[0011] The first neural network unit includes a convolutional layer, a batch normalization layer, a pooling layer, and a Dropout layer; the convolutional layer is used for feature extraction, the batch normalization layer is used to standardize the range of feature data, the pooling layer is used for downsampling, and the Dropout layer is used to update some neuron parameters; the first neural network unit is used to extract features from the filtered signal, the difference signal, and the basic information to obtain feature information;

[0012] The second neural network unit includes a concatenate layer, a flatten layer, and an RNN deep learning network; wherein, the concatenate layer is used for feature fusion processing, and the flatten layer is used to convert the convolutional layer into a fully connected layer; the feature information is fused through the concatenate layer and the flatten layer, and the fused information is passed through the RNN deep learning network to obtain the blood glucose detection result of the object;

[0013] The first output unit is used to output the blood glucose test result;

[0014] Furthermore, in one embodiment of this application, the first light source includes a 1550DFB laser, which is used to generate a light source in the 1550nm band as a measurement light source; the second light source includes a 1310DFB laser, which is used to generate a light source in the 1310nm band as a reference light source.

[0015] Furthermore, in one embodiment of this application, the circuit module further includes: a temperature adjustment unit and a level conversion unit; the temperature adjustment unit is used to adjust the surface temperature of the detection device within a preset temperature range, and the level conversion unit is used to enhance the photoelectric signal generated by the photodetector.

[0016] Furthermore, in one embodiment of this application, the neural network model includes a difference unit, a third neural network unit, and a second output unit; the neural network model is used to predict blood glucose based on the historical blood glucose sequence of an object.

[0017] The difference unit is used to calculate the first and second differences of the historical blood glucose sequence of the object. The first difference is used to characterize the difference between two consecutive adjacent items in the historical blood glucose sequence, and the first difference sequence is obtained through the first difference. The second difference is used to characterize the difference between two consecutive adjacent items in the first difference sequence.

[0018] The third neural network unit is used to predict the object's blood glucose based on the object's historical blood glucose sequence;

[0019] The second output unit is used to output the blood glucose prediction result.

[0020] On the other hand, this application provides a non-invasive blood glucose detection method, applied to the aforementioned non-invasive blood glucose detection system, the method comprising:

[0021] Get basic information about the object;

[0022] The first detector detects the direct photoelectric signal from the first light source as the first electrical signal, and detects the direct photoelectric signal from the second light source as the second electrical signal; the first light source and the second light source are used to emit two optical signals with different wavelengths.

[0023] The third electrical signal of the first light source is detected by the second detector. The third electrical signal is used to characterize the transmitted photoelectric signal or the reflected photoelectric signal of the first light source.

[0024] The fourth electrical signal of the second light source is detected by the third detector. The fourth electrical signal is used to characterize the transmitted photoelectric signal or the reflected photoelectric signal of the second light source.

[0025] The basic information, the first electrical signal, the second electrical signal, the third electrical signal, and the fourth electrical signal are input into a neural network model to detect the blood glucose level of the object, and the blood glucose detection result is obtained.

[0026] Furthermore, the non-invasive blood glucose detection method of this application embodiment further includes:

[0027] A filtered signal is determined based on the first electrical signal and the third electrical signal; the filtered signal is the quotient of the third electrical signal and the first electrical signal.

[0028] The difference signal is determined based on the first electrical signal, the second electrical signal, the third electrical signal, and the fourth electrical signal;

[0029] The difference signal, the filtered signal, and the basic information are processed through a convolutional layer, a batch normalization layer, a pooling layer, and a Dropout layer to obtain feature information; wherein, the convolutional layer is used for feature extraction, the batch normalization layer is used to unify the range of feature data, the pooling layer is used for downsampling, and the Dropout layer is used to update some neuron parameters.

[0030] The feature information is processed through a concatenate layer, a flatten layer, and an RNN deep learning network to obtain the blood glucose detection result of the object; the concatenate layer is used for feature fusion processing, and the flatten layer is used to convert the convolutional layer into a fully connected layer.

[0031] Furthermore, the non-invasive blood glucose detection method of this application embodiment further includes:

[0032] Obtain the historical blood glucose values ​​of the object for a preset number of times as a historical blood glucose sequence;

[0033] Calculate the first difference of the historical blood glucose sequence to obtain the first difference sequence; the first difference is used to characterize the difference between two consecutive adjacent terms in the historical blood glucose sequence;

[0034] Calculate the quadratic difference of the historical blood glucose sequence to obtain a quadratic difference sequence; the quadratic difference is used to characterize the difference between two consecutive adjacent terms in the first difference sequence;

[0035] The first-order difference sequence and the second-order difference sequence are input into the neural network model to obtain the blood glucose prediction result of the object.

[0036] On the other hand, embodiments of this application provide a non-invasive blood glucose detection device, including:

[0037] At least one processor;

[0038] At least one memory for storing at least one program;

[0039] When the at least one program is executed by the at least one processor, the at least one processor implements any of the above-described non-invasive blood glucose detection methods.

[0040] On the other hand, embodiments of this application provide a storage medium storing a processor-executable program, which, when executed by a processor, is used to implement any of the above-described non-invasive blood glucose detection methods.

[0041] This application embodiment inputs the photoelectric signal information of the first and second light sources collected into a neural network model, which can detect blood glucose in the object and improve the accuracy of blood glucose detection. At the same time, the system is simple to operate and improves the user experience. Attached Figure Description

[0042] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the following description is provided with accompanying drawings of the relevant technical solutions in the embodiments of this application or the prior art. It should be understood that the accompanying drawings described below are only for the purpose of clearly illustrating some embodiments of the technical solutions in this application. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0043] Figure 1 A schematic diagram of the structure of one embodiment of the non-invasive blood glucose detection system provided in this application;

[0044] Figure 2A flowchart illustrating one embodiment of the non-invasive blood glucose detection method provided in this application;

[0045] Figure 3 A schematic flowchart of another embodiment of the blood glucose detection method provided in this application;

[0046] Figure 4 A flowchart illustrating one embodiment of the blood glucose prediction method provided in this application;

[0047] Figure 5 This is a schematic diagram of one embodiment of the non-invasive blood glucose testing device provided in this application. Detailed Implementation

[0048] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain this application, and should not be construed as limiting this application. The step numbers in the following embodiments are set only for ease of explanation, and there is no limitation on the order between the steps. The execution order of each step in the embodiments can be adaptively adjusted according to the understanding of those skilled in the art.

[0049] To better understand the system provided in this application, the relevant technologies involved in this application will first be explained:

[0050] A recurrent neural network (RNN) is a type of recurrent neural network that takes sequence data as input, recurses in the direction of sequence evolution, and connects all nodes (recurrent units) in a chain-like manner.

[0051] BatchNormalization layer: refers to the normalization operation performed on the corresponding activation through mini-batch during each stochastic gradient descent.

[0052] PID: The PID algorithm is a linear control method that has proportional, integral, and derivative actions;

[0053] Dropout layer: This is a method used when training a neural network model, to prevent overfitting when there is too little sample data.

[0054] Currently, common methods for blood glucose testing include venous blood sampling and electrochemical glucose testing. Venous blood sampling requires drawing blood to measure blood glucose, limiting its use to hospitals and requiring a long measurement time, making it unsuitable for individuals who need to frequently monitor their blood glucose in daily life. Electrochemical glucose testing, on the other hand, involves applying a voltage to the blood; the blood current increases with the blood glucose concentration. By accurately measuring these weak currents and calculating the corresponding blood glucose concentration based on the relationship between the current value and blood glucose concentration, this method is portable, small in size, and relatively simple to operate. However, it still requires blood sampling, which can cause pain and pose a risk of infection, creating psychological stress for the user. These invasive or minimally invasive methods not only place a physical and psychological burden on the test subject but also increase the risk of unnecessary infection. Therefore, research and application of non-invasive blood glucose testing technologies are becoming increasingly widespread, including methods such as blood replacement therapy, biosensor methods, and energy metabolism conservation methods. Among these methods, the blood substitute method primarily uses human saliva, sweat, and urine. It estimates human blood glucose concentration by calculating the glucose concentration in the substitute, thus establishing a model for analysis. However, experiments have shown a weak correlation between the substitute and blood, requiring further investigation. The biosensor method uses external or implantable subcutaneous sensors to measure physical quantities in the human body, which show a strong correlation with blood glucose levels. However, this method may cause harmful radiation to human skin tissue, thus limiting its widespread application. The energy conservation method is based on the energy conservation principle in daily metabolism, where the heat and oxygen consumption generated during glucose consumption are related to blood glucose concentration. However, actual clinical studies require measuring parameters such as body temperature and environmental temperature and humidity, necessitating the use of numerous sensors. Therefore, it is susceptible to errors in the specifications and accuracy of each sensor, requiring further research and improvement for non-invasive detection methods based on the energy conservation principle.

[0055] Therefore, it is evident that current blood glucose testing systems or methods have the following problems:

[0056] (1) First, existing invasive and minimally invasive blood glucose testing methods take a long time to measure, cause pain to patients and pose a risk of infection with other diseases, and cause psychological stress to users during measurement; in addition, the operation is relatively complicated and highly dependent on the operator's skill level, making it unsuitable for long-term and repeated use by individuals.

[0057] (2) Non-invasive blood glucose detection technologies such as blood substitute method, biosensor method, and energy metabolism method have low feature correlation, require a large number of physical sensor parameters, and may cause damage to human skin, making them unsuitable for application.

[0058] (3) The measurement signals of blood glucose concentration obtained by various non-invasive blood glucose detection methods are easily affected by many external factors, resulting in a large error in the blood glucose value measured by the system.

[0059] The non-invasive blood glucose detection system and method according to the embodiments of this application are described in detail below with reference to the accompanying drawings. First, a non-invasive blood glucose detection system according to the embodiments of this application is described with reference to the accompanying drawings.

[0060] Figure 1 This is a schematic diagram of a non-invasive blood glucose detection system according to an embodiment of this application. The system specifically includes:

[0061] The system comprises an optical path module, a circuit module, an interaction module, and a control module; one end of the optical path module is connected to the control module via the circuit module, the other end of the optical path module is connected to the control module, and the control module is connected to the interaction module.

[0062] The optical path module includes a light emitter, a photodetector, and a detection device. The light emitter includes a first light source and a second light source, which emit two light signals of different wavelengths. The detection device is positioned near a preset collection point on the object to perform blood glucose detection. The photodetector includes a first detector, a second detector, and a third detector. The first detector detects the direct photoelectric signals from the first and second light sources, the second detector detects the transmitted or reflected photoelectric signals from the first light source, and the third detector detects the transmitted or reflected photoelectric signals from the second light source. The optical path module, after the light emitted by the light emitter illuminates the preset collection point on the object, determines the photoelectric signal information through photoelectric detection.

[0063] The circuit module includes a time-division multiplexing unit; the time-division multiplexing unit is used to realize the time-division multiplexing of the first light source and the second light source;

[0064] The interaction module is used to receive basic information about the object;

[0065] The control module includes a neural network model, which is used to detect blood glucose levels in the object based on the photoelectric signal information and the basic information.

[0066] Specifically, the non-invasive blood glucose detection system proposed in this application refers to... Figure 1As shown, the optical path module includes a light emitter, a photodetector, and a detection device. In some possible implementations, the light emitter may include a 1550nm laser source, denoted as the first light source; the light emitter may include a 1310nm laser source, denoted as the second light source. The optical path module is used to determine photoelectric signal information by photodetection after the light emitted by the light emitter illuminates a preset collection area of ​​the object.

[0067] In some possible implementations, the circuit module further includes a constant current drive unit, a temperature regulation unit, a current / voltage (I / V) conversion circuit, an amplification and filtering circuit, and a level conversion circuit to adjust the operating state of the light emitter and the signal reception of the photodetector. In some possible implementations, the control module may also include a microprocessor unit for controlling the operating state of the circuit module and receiving photoelectric signal information from the optical path module. Similarly, the interaction module can also be used to display blood glucose test results and prediction results to the user.

[0068] In some possible implementations, human blood is a red, opaque liquid produced by the heart and circulating in the blood vessels. It is mainly composed of plasma and blood cells. Plasma contains glucose, inorganic salts, and a large amount of water. The glucose concentration in the blood refers to blood glucose concentration. The chemical formula of glucose contains multiple methyl (OH) and hydroxyl (CH) groups. These functional groups are hydrogen-containing functional groups that absorb near-infrared spectroscopy strongly. Typically, in the wavelength range of 1000-1400 nm, the absorption of glucose molecules is small, but the absorption of water molecules is large; in the wavelength range of 1400-1800 nm, the absorption of glucose molecules is strong, but the absorption of water molecules is weak. Therefore, the system will select appropriate wavelengths in the two bands as the measurement light source and reference light source, respectively. The reason for choosing a reference light source is that the water molecule content in blood is relatively high, so the influence of water molecules on the system measurement should be avoided when selecting the wavelength. To obtain the best detection effect, a suitable detection site needs to be selected, which should meet the following requirements: (1) It should be an exposed part of the human body for easy detection; (2) It should be a part with less individual differences among users; (3) It should be a measurement environment with less external interference; (4) It should be a detection site with rich capillaries and less content of other tissue components. In addition, the selected site should avoid interference from human body surface factors, such as body temperature and sweat. Considering the above factors, the detection site (i.e., the preset collection site) can be selected, such as the finger or ear. Near-infrared light of a fixed wavelength is used to irradiate one end of the finger, and a highly sensitive photodetector is placed at the other end of the finger. The near-infrared light is received by the photodetector after passing through the blood in the skin. Since the absorption of light by bone tissue in the skin is always constant, while the blood volume changes pulsatially, that is, when the heart contracts, the peripheral blood volume is the largest, and the light absorption is the largest, so the detected light intensity is the smallest; when the heart relaxes, the peripheral blood volume is the smallest, and the light absorption is the smallest, so the detected light intensity is the largest. Therefore, the intensity of light detected by the light source changes pulsatingly, just like the blood volume. Furthermore, the spectrum detected at specific wavelengths carries a wealth of detection information. The near-infrared-based non-invasive blood glucose monitoring system truly achieves non-invasive detection during measurement, requiring only the placement of a finger in a fixed position. It uses a pre-established neural network model for prediction, making operation simple and analysis fast and straightforward. The system's circuit and optical modules are clearly defined, and the device is portable, enabling real-time and repetitive use, making it suitable for daily blood glucose monitoring.

[0069] Optionally, in the non-invasive blood glucose detection system of this application embodiment, the neural network model includes a data input unit, a preprocessing unit, a first neural network unit, a second neural network unit, and a first output unit;

[0070] The data input unit is used to input the photoelectric signal information and the basic information, and the preprocessing unit is used to filter and normalize the photoelectric signal information to obtain the filtered signal and the difference signal.

[0071] The first neural network unit includes a convolutional layer, a batch normalization layer, a pooling layer, and a Dropout layer; the convolutional layer is used for feature extraction, the batch normalization layer is used to standardize the range of feature data, the pooling layer is used for downsampling, and the Dropout layer is used to update some neuron parameters; the first neural network unit is used to extract features from the filtered signal, the difference signal, and the basic information to obtain feature information;

[0072] The second neural network unit includes a concatenate layer, a flatten layer, and an RNN deep learning network; wherein, the concatenate layer is used for feature fusion processing, and the flatten layer is used to convert the convolutional layer into a fully connected layer; the feature information is fused through the concatenate layer and the flatten layer, and the fused information is passed through the RNN deep learning network to obtain the blood glucose detection result of the object;

[0073] The first output unit is used to output the blood glucose test result.

[0074] Specifically, this application uses a neural network model to detect blood glucose levels in a target. In related technologies, near-infrared spectroscopy is used to irradiate areas of the human body rich in capillaries and with fewer other tissues. Chemometric methods are used to obtain the absorbance of glucose in the blood, and a mathematical model is established between this absorbance and blood glucose concentration to measure the blood glucose concentration. When a light source shines on the fingertip, the transmitted photoelectric signal has two main components: a constant component (DC), which mainly reflects the absorption of light by venous blood, muscles, and bones; and a pulsating component (AC), which is a periodically pulsating waveform synchronized with the pulse rate. The superposition of these two components reflects the absorption of light by blood glucose in the blood. The light intensity transmitted through the fingertip is denoted as:

[0075]

[0076] Where ε0, c0, and L are the total absorbance, concentration of the light-absorbing substance, and optical path length in the tissue, respectively; ε1 and c1 are the absorption coefficients and concentrations of glucose in arterial blood, respectively; ε2 and c2 are the absorption coefficients and concentrations of the reference substance in arterial blood, respectively; and I0 is the input light intensity. During cardiac diastole and systole, the blood volume in the blood vessels changes, which leads to a change in the optical path length in the arterial blood, increasing from the original L to L + ΔL, i.e.:

[0077]

[0078] Dividing formula (2) by (1), we get:

[0079] Since ΔL / I << 1, then we have If two different wavelengths, λ1 and λ2, are used in the experiment, then:

[0080]

[0081]

[0082] Subtracting equation (4) from equation (3) gives: when Then we have: make Let R be the denoted value, then we have Therefore, the ratio R of red light and infrared light and the blood glucose value c1 can be obtained based on Lambert-Beer quantification.

[0083] However, the above methods estimate blood glucose levels using mathematical models, resulting in low accuracy. Therefore, this application proposes a method for estimating blood glucose levels based on artificial intelligence. The specific system, when applied, executes the following steps:

[0084] When the system powers on and the user clicks the "Start Test" button on the system screen, basic information about the user, such as age, height, weight, and last meal time, is collected and stored in an array. The embedded microprocessor unit uses timer control via GPIO to automatically adjust the 1550nm laser to operate for 15 seconds at regular intervals. A dual-channel ADC is used to collect transmitted and direct near-infrared light data in the 1550nm band. SPI communication is used to operate the built-in SD card and store the currently collected data. Wireless communication is used to transmit the collected dual-channel data to a remote host. Next, the embedded microprocessor unit uses GPIO to automatically control the 1310nm laser to operate for 15 seconds at regular intervals. Similarly, a dual-channel ADC is used to collect near-infrared light data in this band, which is stored on the SD card and transmitted wirelessly to the remote host. In addition, the system is equipped with indicator lights for the 1550nm and 1310nm data acquisition processes to indicate the current operating status. The two laser light sources are time-division multiplexed, and the two data sources are acquired in real time using a dual-channel ADC and stored on the system's SD card, and then transmitted wirelessly.

[0085] After data collection is complete, the system loads a pre-trained RNN neural network model based on multi-data fusion. The model takes the stored basic information of the object, the collected 1550 nm near-infrared direct and transmitted light data, and the collected 1310 nm near-infrared direct and transmitted light data as input. The loaded model estimates the human blood glucose level based on the input data and displays it on the system screen. Furthermore, after 10 consecutive measurements of an individual, clicking the blood glucose trend prediction button will load the measured blood glucose value sequence based on the RNN neural network to depict the trend of the object's blood glucose level changes, thus completing the human blood glucose monitoring function. It should be noted that the above process is an exemplary example and does not constitute a specific limitation on the first power source, the second power source, the number of parameters in the historical blood glucose sequence, or the type of light signal received by the detector in this application.

[0086] Reference Figure 3 As shown, the processing procedure of the neural network model can be as follows: The system transmits the collected 1550nm direct and transmitted signals, 1310nm direct and transmitted signals, and basic information about the object to the host via wireless communication. The host uses a pre-built RNN deep learning neural network based on multi-data fusion for model training. Let the 1550nm transmitted and direct signals be x1 and x2, and the 1310nm transmitted and direct signals be y1 and y2, respectively. Then, (x1-y1) / (x2-y2) is calculated to obtain the signal with background noise removed, as well as the original signal (x1 / x2). The signal with background noise removed, i.e., the original signal, undergoes data preprocessing operations such as filtering and normalization. The filtered difference signal z1 and the filtered signal z2 are multiplied by weights w1 and w2, respectively, to obtain Z1' and Z2'. Both datasets are processed through a network of four different types: convolutional layer, batch normalization layer, pooling layer, and dropout layer, to obtain the network outputs Tensor_x and Tensor_y. In addition, basic information about the collected subjects, such as age, height, weight, and the time of their last meal, is used as another input. This information is then processed through a convolutional layer-batch normalization layer-pooling layer-Dropout layer network to obtain the output Tensor_z. The Tensor_x, Tensor_y, and Tensor_z outputs are fused through a concatenation layer-Flatten layer. The fused data output is then fed into an RNN deep learning neural network for model training. The data output is processed through four different types of Dense-batch normalization-Dropout layers to obtain the final blood glucose prediction output value. Furthermore, referring to... Figure 4As shown, when an individual user measures their blood glucose level more than 10 times consecutively, the system uses the measured blood glucose value sequence as input to the blood glucose value change trend model, calculates the first difference y1 and the second difference y2 of the sequence, and trains the model through an RNN-Dense layer. After training, prediction, and calculation, the model will obtain the blood glucose prediction value for the next moment, which will be used as the input for the blood glucose prediction at the next moment. This realizes the prediction of the trend of blood glucose value change and realizes the blood glucose monitoring function of the system.

[0087] Optionally, in the non-invasive blood glucose detection system of this application embodiment, the first light source includes a 1550DFB laser, which is used to generate a light source in the 1550nm band as a measurement light source; the second light source includes a 1310DFB laser, which is used to generate a light source in the 1310nm band as a reference light source.

[0088] Specifically, in the optical system setup, two DFB lasers of different wavelengths are used as the light source. DFB lasers are characterized by high output power, good light source stability, and good monochromaticity. With proper design of the peripheral constant current drive circuit and temperature control circuit, the laser diode can operate in a stable and optimal state. Two light sources, 1550nm and 1310nm, are selected as the system's measurement and reference light sources. The 1550nm measurement light source is used to detect glucose molecules in the blood, while the 1310nm reference light source is used to measure water molecules in the blood, thus removing background noise during blood glucose detection. The two laser source signals are connected to the input of a coupler. One of the coupler's two outputs is connected to a photodetector with a collimator, and the other is used to illuminate a finger. The system uses a photodetector with low dark current and high sensitivity and responsivity to receive both the direct near-infrared light signal and the near-infrared light signal transmitted through the finger. Because the current output by the photodetector is relatively weak, subsequent signal conditioning circuits, such as a two-stage amplification and filtering circuit and a level conversion circuit, need to be designed.

[0089] Optionally, in the non-invasive blood glucose detection system of this application embodiment, the circuit module further includes: a temperature adjustment unit and a level conversion unit; the temperature adjustment unit is used to adjust the surface temperature of the detection device within a preset temperature range, and the level conversion unit is used to enhance the photoelectric signal generated by the photodetector.

[0090] Specifically, in the hardware system design, a constant current drive circuit based on FP7103 is adopted, and the constant current regulation formula of the circuit output is: I F =V FB / R, where V FBThe fixed output is 0.25V, so adjusting R provides a constant output current. A temperature control circuit based on the MAX1978 core is used, with an external thermistor bridge circuit. A thermistor is used to monitor the laser surface temperature in real time. If temperature fluctuations are large, a designed H-bridge circuit and PID control circuit are used to adjust the circuit current, thereby regulating the TEC heating or cooling power and maintaining the laser at a constant surface temperature. In some possible implementations, time-division multiplexing is used to allow the 1550 measurement light source and the 1310 reference light source to operate independently, avoiding mutual interference. Specifically, a transistor-driven light source circuit is used, and the drive I / O is connected to the GPIO control terminal of the embedded microprocessor unit. The microprocessor unit controls the high and low levels of the GPIO output and uses a timer, thus achieving time-division multiplexing of the two light sources. An I / V conversion circuit based on the OPA131 photovoltaic mode is designed. When near-infrared light shines on the surface of the photodetector, the photodiode generates a current proportional to the intensity of the irradiated light, but this current is relatively weak. The current passes through the feedback resistor RF and is converted into voltage. Because noise such as 50Hz power frequency interference exists during signal acquisition, a two-stage amplifier circuit was designed, and a low-pass filter circuit was also introduced. This retains the effective low-frequency signal, filters out high-frequency noise, and amplifies the effective signal. Since the laser detector receiving circuit uses multiple power supply types, a 5V regulated output circuit based on the RT9163 and a ±5V output circuit based on the MAX860 are both available. Furthermore, a regulated power supply is designed and provides multiple power interfaces to facilitate power supply usage within the system.

[0091] Optionally, in the non-invasive blood glucose detection system of this application embodiment, the neural network model includes a differential unit, a third neural network unit, and a second output unit; the neural network model is used to predict blood glucose based on the historical blood glucose sequence of the object.

[0092] The difference unit is used to calculate the first and second differences of the historical blood glucose sequence of the object. The first difference is used to characterize the difference between two consecutive adjacent items in the historical blood glucose sequence, and the first difference sequence is obtained through the first difference. The second difference is used to characterize the difference between two consecutive adjacent items in the first difference sequence.

[0093] The third neural network unit is used to predict the object's blood glucose based on the object's historical blood glucose sequence;

[0094] The second output unit is used to output the blood glucose prediction result.

[0095] In summary, the non-invasive blood glucose monitoring system proposed in this application constructs an optical path system with two DFB lasers with wavelengths of p1=1550nm and p2=1310nm, and three detectors. It includes a constant current source drive circuit, a TEC temperature control circuit based on PID regulation, a signal amplification and filtering circuit, and a level conversion circuit. The system utilizes a multi-channel ADC for real-time multiplexing of the two DFB lasers controlled by an embedded microprocessor unit, and features multi-channel ADC acquisition, storage, and wireless transmission capabilities. To further improve the model's blood glucose prediction performance, the system incorporates basic information about the target individual and uses an RNN network model based on multi-data fusion to predict individual blood glucose levels, while also providing trend prediction of individual blood glucose concentrations and individual blood glucose monitoring alerts. Compared to invasive or minimally invasive blood glucose monitoring technologies, this system eliminates the risk of pain and infection for the test subject. Compared to current non-invasive research methods such as blood replacement therapy, biosensor methods, and energy metabolism conservation methods, the system, through artificial intelligence technology, offers highly reliable and accurate test results. Meanwhile, the system is simple and quick to operate, requiring only fingertip placement for data collection, making it suitable for continuous and repeated blood glucose testing by individuals, and it is safe and reliable. Furthermore, the system utilizes the system board's blood glucose concentration prediction and individual blood glucose concentration change trend functions, providing blood glucose monitoring capabilities. In addition, it features remote model training, testing, blood glucose prediction, blood glucose concentration change trend assessment, and blood glucose monitoring functions.

[0096] Next, the non-invasive blood glucose detection method proposed according to the embodiments of this application will be described with reference to the accompanying drawings.

[0097] Reference Figure 2 This application provides a non-invasive blood glucose detection method. This method can be applied to a terminal, a server, or software running on either a terminal or server. The terminal can be a tablet, laptop, desktop computer, etc., but is not limited to these. The server can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The non-invasive blood glucose detection method described in this application, applied to the aforementioned non-invasive blood glucose detection system, mainly includes the following steps:

[0098] S210: Obtain basic information about the object;

[0099] S220: The first detector detects the direct photoelectric signal from the first light source as the first electrical signal, and detects the direct photoelectric signal from the second light source as the second electrical signal; the first light source and the second light source are used to emit two optical signals with different wavelengths;

[0100] S230: The third electrical signal of the first light source is detected by the second detector, and the third electrical signal is used to characterize the transmitted photoelectric signal or the reflected photoelectric signal of the first light source;

[0101] S240: The fourth electrical signal of the second light source is detected by the third detector, the fourth electrical signal being used to characterize the transmitted photoelectric signal or reflected photoelectric signal of the second light source;

[0102] S250: Input the basic information, the first electrical signal, the second electrical signal, the third electrical signal and the fourth electrical signal into the neural network model to detect the blood glucose of the object and obtain the blood glucose detection result;

[0103] Optionally, the non-invasive blood glucose detection method in this application embodiment further includes: determining a filtered signal based on the first electrical signal and the third electrical signal; the filtered signal is the quotient of the third electrical signal and the first electrical signal;

[0104] The difference signal is determined based on the first electrical signal, the second electrical signal, the third electrical signal, and the fourth electrical signal;

[0105] The difference signal, the filtered signal, and the basic information are processed through a convolutional layer, a batch normalization layer, a pooling layer, and a Dropout layer to obtain feature information; wherein, the convolutional layer is used for feature extraction, the batch normalization layer is used to unify the range of feature data, the pooling layer is used for downsampling, and the Dropout layer is used to update some neuron parameters.

[0106] The feature information is processed through a concatenate layer, a flatten layer, and an RNN deep learning network to obtain the blood glucose detection result of the object; the concatenate layer is used for feature fusion processing, and the flatten layer is used to convert the convolutional layer into a fully connected layer.

[0107] Optionally, the non-invasive blood glucose detection method in this application embodiment further includes: obtaining the historical blood glucose values ​​of the object a preset number of times as a historical blood glucose sequence;

[0108] Calculate the first difference of the historical blood glucose sequence to obtain the first difference sequence; the first difference is used to characterize the difference between two consecutive adjacent terms in the historical blood glucose sequence;

[0109] Calculate the quadratic difference of the historical blood glucose sequence to obtain a quadratic difference sequence; the quadratic difference is used to characterize the difference between two consecutive adjacent terms in the first difference sequence;

[0110] The first-order difference sequence and the second-order difference sequence are input into the neural network model to obtain the blood glucose prediction result of the object.

[0111] It is evident that the content of the above system embodiments is applicable to this method embodiment. The specific functions implemented in this method embodiment are the same as those in the above system embodiments, and the beneficial effects achieved are also the same as those achieved in the above system embodiments.

[0112] Reference Figure 5 This application provides a non-invasive blood glucose testing device, comprising:

[0113] At least one processor 510;

[0114] At least one memory 520 is used to store at least one program;

[0115] When the at least one program is executed by the at least one processor 510, the at least one processor 510 implements the non-invasive blood glucose detection method.

[0116] Similarly, the content of the above method embodiments is applicable to this device embodiment. The specific functions implemented by this device embodiment are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0117] This application embodiment also provides a computer-readable storage medium storing a program executable by a processor 510, which, when executed by the processor 510, is used to perform the above-described non-invasive blood glucose detection method.

[0118] Similarly, the content of the above method embodiments is applicable to this storage medium embodiment. The specific functions implemented in this storage medium embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0119] In some alternative embodiments, the functions / operations mentioned in the block diagrams may not occur in the order shown in the operation diagrams. For example, depending on the functions / operations involved, two consecutively shown blocks may actually be executed substantially simultaneously, or the blocks may sometimes be executed in reverse order. Furthermore, the embodiments presented and described in the flowcharts of this application are provided by way of example to provide a more comprehensive understanding of the technology. The disclosed methods are not limited to the operations and logic flows presented herein. Alternative embodiments are contemplated in which the order of various operations is changed and sub-operations described as part of a larger operation are executed independently.

[0120] Furthermore, although this application is described in the context of functional modules, it should be understood that, unless otherwise stated to the contrary, one or more of the functions and / or features may be integrated into a single physical device and / or software module, or one or more functions and / or features may be implemented in a separate physical device or software module. It is also understood that a detailed discussion of the actual implementation of each module is unnecessary for understanding this application. Rather, given the properties, functions, and internal relationships of the various functional modules in the apparatus disclosed herein, the actual implementation of the module will be understood within the scope of conventional technology for an engineer. Therefore, those skilled in the art can implement the application set forth in the claims using ordinary techniques without excessive experimentation. It is also understood that the specific concepts disclosed are merely illustrative and not intended to limit the scope of this application, which is determined by the full scope of the appended claims and their equivalents.

[0121] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several programs to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0122] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequential list of executable programs for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, a program execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can retrieve and execute a program from or in conjunction with such a program execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can mean any means that can contain, store, communicate, propagate, or transmit a program for use by or in conjunction with a program execution system, apparatus, or device.

[0123] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0124] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable program execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0125] In the foregoing description of this specification, the references to terms such as "one embodiment," "another embodiment," or "some embodiments," etc., indicate that a specific feature, structure, material, or characteristic described in connection with an embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0126] Although embodiments of this application have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of this application, the scope of which is defined by the claims and their equivalents.

[0127] The above is a detailed description of the preferred embodiments of this application, but this application is not limited to the embodiments described. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of this application, and these equivalent modifications or substitutions are all included within the scope defined by the claims of this application.

Claims

1. A non-invasive blood glucose detection system, characterized by, The detection system includes: an optical path module, a circuit module, an interaction module, and a control module; one end of the optical path module is connected to the control module through the circuit module, the other end of the optical path module is connected to the control module, and the control module is connected to the interaction module; The optical path module includes a light emitter, a photodetector, and a detection device. The light emitter includes a first light source and a second light source. The first light source is a 1550nm DFB laser, and the second light source is a 1310nm DFB laser. The first and second light sources emit two light signals with different wavelengths. The detection device is used to approach a preset collection point on the object for blood glucose detection. The photodetector includes a first detector, a second detector, and a third detector. The first detector detects the direct photoelectric signals from the first and second light sources. The second detector detects the transmitted or reflected photoelectric signals from the first light source. The third detector detects the transmitted or reflected photoelectric signals from the second light source. The optical path module uses the light emitted by the light emitter to illuminate the preset collection point on the object and then uses photoelectric detection to determine the photoelectric signal information. The circuit module includes a time-division multiplexing unit, a TEC temperature regulation unit, and a PID control unit; the time-division multiplexing unit is used to realize the time-division multiplexing of the first light source and the second light source; the TEC temperature regulation unit, in conjunction with the PID control unit, is used to regulate the laser operating temperature and stabilize the light source operating point; The interaction module is used to receive basic information about the object; The control module includes an embedded acquisition unit, a wireless transmission unit, a remote host model training and inference unit, and a neural network model. The embedded acquisition unit is used to acquire and store photoelectric signal information, and transmits the data to the remote host through the wireless transmission unit, whereby the remote host completes model training and blood glucose prediction. The neural network model is used to detect the blood glucose of the object based on the photoelectric signal information and the basic information, and to predict the blood glucose based on the object's historical blood glucose sequence. The neural network model includes a data input unit, a preprocessing unit, a first neural network unit, a second neural network unit, a first output unit, a differential unit, a third neural network unit, and a second output unit. The data input unit is used to input the photoelectric signal information and the basic information; the preprocessing unit is used to filter and normalize the photoelectric signal information to obtain a filtered signal and a difference signal; the first neural network unit includes a convolutional layer, a batch normalization layer, a pooling layer, and a Dropout layer; the convolutional layer is used for feature extraction, the batch normalization layer is used to unify the range of feature data, the pooling layer is used for downsampling, and the Dropout layer is used to update some neuron parameters; the first neural network unit is used to extract features from the filtered signal, the difference signal, and the basic information to obtain feature information; the second neural network unit includes a concatenate layer, a flatten layer, and an RNN deep learning network; the concatenate layer is used for feature fusion processing, and the flatten layer is used to convert the convolutional layer to a fully connected layer; the feature information is fused through the concatenate layer and the flatten layer, and the fused information is passed through the RNN deep learning network to obtain the blood glucose detection result of the object; the first output unit is used to output the blood glucose detection result. The difference unit is used to calculate the first and second differences of the object's historical blood glucose sequence. The first difference is used to characterize the difference between two consecutive adjacent items in the historical blood glucose sequence, and a first difference sequence is obtained through the first difference. The second difference is used to characterize the difference between two consecutive adjacent items in the first difference sequence. The third neural network unit is used to predict the object's blood glucose based on the object's historical blood glucose sequence. The second output unit is used to output the blood glucose prediction result.

2. The non-invasive blood glucose detection system according to claim 1, characterized in that, The first light source includes a 1550DFB laser, which is used to generate a light source in the 1550nm band as a measurement light source; the second light source includes a 1310DFB laser, which is used to generate a light source in the 1310nm band as a reference light source.

3. The non-invasive blood glucose detection system according to claim 1, characterized in that, The circuit module further includes a temperature adjustment unit and a level conversion unit; the temperature adjustment unit is used to adjust the surface temperature of the detection device within a preset temperature range, and the level conversion unit is used to enhance the photoelectric signal generated by the photodetector.

4. A non-invasive blood glucose detection method, characterized in that, The method, applied to the non-invasive blood glucose detection system as described in claim 1, comprises: Get basic information about the object; The first detector detects the direct photoelectric signal from the first light source as the first electrical signal, and detects the direct photoelectric signal from the second light source as the second electrical signal; the first light source and the second light source are used to emit two optical signals with different wavelengths. The third electrical signal of the first light source is detected by the second detector. The third electrical signal is used to characterize the transmitted photoelectric signal or the reflected photoelectric signal of the first light source. The fourth electrical signal of the second light source is detected by the third detector. The fourth electrical signal is used to characterize the transmitted photoelectric signal or the reflected photoelectric signal of the second light source. A filtered signal is determined based on the first electrical signal and the third electrical signal; the filtered signal is the quotient of the third electrical signal and the first electrical signal; a difference signal is determined based on the first electrical signal, the second electrical signal, the third electrical signal, and the fourth electrical signal. Based on the Lambert-Beer law, dual-wavelength near-infrared photophysiological modeling was performed on the first, second, third, and fourth electrical signals to derive the quantitative relationship between the intensity difference of the dual-wavelength light and blood glucose concentration. The difference signal, the filtered signal, and the basic information are processed through a convolutional layer, a batch normalization layer, a pooling layer, and a Dropout layer to obtain feature information; wherein, the convolutional layer is used for feature extraction, the batch normalization layer is used to unify the range of feature data, the pooling layer is used for downsampling, and the Dropout layer is used to update some neuron parameters. The feature information is processed through a concatenate layer, a flatten layer, and an RNN deep learning network to obtain the blood glucose detection result of the object; the concatenate layer is used for feature fusion processing, and the flatten layer is used to convert the convolutional layer into a fully connected layer; Ten consecutive historical blood glucose values ​​of an object are obtained as a historical blood glucose sequence. The first and second differences of the historical blood glucose sequence are calculated. The first and second difference sequences are input into a neural network model to obtain the object's blood glucose trend prediction result.

5. The non-invasive blood glucose detection method according to claim 4, characterized in that, The method further includes: Obtain the historical blood glucose values ​​of the object for a preset number of times as a historical blood glucose sequence; Calculate the first difference of the historical blood glucose sequence to obtain the first difference sequence; the first difference is used to characterize the difference between two consecutive adjacent terms in the historical blood glucose sequence; Calculate the quadratic difference of the historical blood glucose sequence to obtain a quadratic difference sequence; the quadratic difference is used to characterize the difference between two consecutive adjacent terms in the first difference sequence; The first-order difference sequence and the second-order difference sequence are input into the neural network model to obtain the blood glucose prediction result of the object.

6. A non-invasive blood glucose testing device, characterized in that, include: At least one processor; At least one memory for storing at least one program; When the at least one program is executed by the at least one processor, the at least one processor implements the non-invasive blood glucose detection method as described in any one of claims 4-5.

7. A computer-readable storage medium storing a processor-executable program, characterized in that: The processor-executable program, when executed by the processor, is used to implement the non-invasive blood glucose detection method as described in any one of claims 4-5.

Images